_Democracy.qmd

# Data

-   Top 1% Data from WID.World
-   Democracy and other Data from ANRR

## Imports

```{r}
library(tidyverse)
library(wid)
library(countrycode)
library(plm)
```

## Democracy Data

```{r}
DDCG <- haven::read_dta("data/DDCGdata.dta")
```

Clean the Data

```{r}
DDCG <- DDCG %>%
  dplyr::select("country_name","wbcode","dem","y", "yeardem", "year", "demreg")
```

Get the Country Codes from the DDCG to download the respective WID Data

```{r}
iso3 <- DDCG$wbcode %>% unique()
iso2 <- countrycode(iso3, 
            origin = "iso3c", destination = "iso2c", 
            custom_match = c(
              "NAU" = "NR",
              "ROM" = "RO",
              "SER" = "RS",
              "SIN" = "SG",
              "TAW" = "TW",
              "UAE" = "AE",
              "ZAR" = "CD"
            ))
```

## 1% Data WID

```{r}
WID <- download_wid(
  indicators = "sptinc", # Shares of pre-tax national income
  areas = iso2, # In the United States
  years = 1960:2010, # Time period: 2010-2015
  perc = "p99p100", # Top 1% only
  pop = "j", # Population
  include_extrapolations = TRUE, # Include extrapolations
)
```

Convert WID to ISO3 Country Codes

```{r}
#| cache: true
WID <- WID %>%
  mutate(iso3 = countrycode(country, origin = "iso2c", destination = "iso3c", 
                            custom_match = c(
                              "NR" = "NAU", 
                              "RO" = "ROM", 
                              "RS" = "SER", 
                              "SG" = "SIN", 
                              "TW" = "TAW", 
                              "AE" = "UAE", 
                              "CD" = "ZAR")
                            )) %>%
  rename(top = value) %>%
  #only take values where variable = sptinc992j
  filter(variable == "sptinc992j")
```

Merge the two Data Sets

```{r}
data <- DDCG %>%
  left_join(WID, by = c("wbcode" = "iso3", "year" = "year")) %>%
  select(country_name, wbcode, year, dem, y, yeardem, top, demreg) %>%
  mutate(ltop = log(top)) #logarithmize
```

## 1% Fixed Effects Models

Build general model

```{r}
pdf <- pdata.frame(data, index = c("wbcode", "year"))
```

### Simple Model with Log

```{r}
model <- plm(
  ltop ~ dem,
  data = pdf,
  model = "within",
  effect = "twoways"
  )
summary(model)
```

-   0.07\*100 = 7% change in share= lower share of top 1% rises
-   statistically significant
-   but R2 is very low (not good at explaining)
    -   but R2 is in Fix Ef Model not that good at explaining?

### Model with 2 Lags

```{r}
model2 <- plm(
  ltop ~ dem + plm::lag(ltop, 1:2),
  data = pdf,
  model = "within",
  effect = "twoways"
  )
summary(model2)
```

-   coefficient not significant!
-   R2 very good because of the lag...

## IV Model

```{r}
iv_model_1 <- plm(
    y ~ dem  | 
    plm::lag(demreg,1:4),
    data=pdf,
    effect = "twoways",
    )
summary(iv_model_1)
```

= nope

## shorter time frame

with years from 1980

```{r}
data_t2 <- data %>%
  filter(year >= 1980)
```

```{r}
pdf_t2 <- pdata.frame(data_t2, index = c("wbcode", "year"))

model_t2 <- plm(
  ltop ~ dem,
  data = pdf_t2,
  model = "within",
  effect = "twoways"
  )
summary(model_t2)
```

= nope, similar to normal

## Wealth instead of Income

WID Code = shweal992j

```{r}
WID_wealth <- download_wid(
  indicators = "shweal", # Shares of pre-tax national income
  areas = iso2, # In the United States
  years = 1960:2010, # Time period: 2010-2015
  perc = "p99p100", # Top 1% only
  pop = "j", # Population
  include_extrapolations = TRUE, # Include extrapolations
)
```

Fix

```{r}
WID_wealth <- WID_wealth %>%
  mutate(iso3 = countrycode(country, origin = "iso2c", destination = "iso3c", 
                            custom_match = c(
                              "NR" = "NAU", 
                              "RO" = "ROM", 
                              "RS" = "SER", 
                              "SG" = "SIN", 
                              "TW" = "TAW", 
                              "AE" = "UAE", 
                              "CD" = "ZAR")
                            )) %>%
  rename(top = value) %>%
  #only take values where variable = sptinc992j
  filter(variable == "shweal992j")
```

```{r}
data_wealth <- DDCG %>%
  left_join(WID_wealth, by = c("wbcode" = "iso3", "year" = "year")) %>%
  select(country_name, wbcode, year, dem, y, yeardem, top, demreg) %>%
  mutate(ltop = log(top)) #logarithmize
```

```{r}
pdf_wealth <- pdata.frame(data_wealth, index = c("wbcode", "year"))
```

Model

```{r}
model_wealth <- plm(
  ltop ~ dem,
  data = pdf_wealth,
  model = "within",
  effect = "twoways"
  )
summary(model_wealth)
```

## Tests with other Income Shares

### 10%

```{r}
WID10 <- download_wid(
  indicators = "sptinc", # Shares of pre-tax national income
  areas = iso2, # In the United States
  years = 1960:2010, # Time period: 2010-2015
  perc = "p90p100", # Top 1% only
  pop = "j", # Population
  include_extrapolations = TRUE, # Include extrapolations
)
```

Convert WID to ISO3 Country Codes

```{r}
#| cache: true
WID10 <- WID10 %>%
  mutate(iso3 = countrycode(country, origin = "iso2c", destination = "iso3c", 
                            custom_match = c(
                              "NR" = "NAU", 
                              "RO" = "ROM", 
                              "RS" = "SER", 
                              "SG" = "SIN", 
                              "TW" = "TAW", 
                              "AE" = "UAE", 
                              "CD" = "ZAR")
                            )) %>%
  rename(top = value) %>%
  #only take values where variable = sptinc992j
  filter(variable == "sptinc992j")
```

Merge the two Data Sets

```{r}
data10 <- DDCG %>%
  left_join(WID10, by = c("wbcode" = "iso3", "year" = "year")) %>%
  select(country_name, wbcode, year, dem, y, yeardem, top)
pdf10 <- pdata.frame(data10, index = c("wbcode", "year"))
```

Simple Model

```{r}
model10_1 <- plm(
  log(top) ~ dem,
  data = pdf10,
  model = "within",
  effect = "twoways"
  )
summary(model10_1)
```

= similar to before

### 50%

```{r}
WID50 <- download_wid(
  indicators = "sptinc", # Shares of pre-tax national income
  areas = iso2, # In the United States
  years = 1960:2010, # Time period: 2010-2015
  perc = "p50p100", # Top 1% only
  pop = "j", # Population
  include_extrapolations = TRUE, # Include extrapolations
)
```

Convert WID to ISO3 Country Codes

```{r}
WID50 <- WID50 %>%
  mutate(iso3 = countrycode(country, origin = "iso2c", destination = "iso3c", 
                            custom_match = c(
                              "NR" = "NAU", 
                              "RO" = "ROM", 
                              "RS" = "SER", 
                              "SG" = "SIN", 
                              "TW" = "TAW", 
                              "AE" = "UAE", 
                              "CD" = "ZAR")
                            )) %>%
  rename(top = value) %>%
  #only take values where variable = sptinc992j
  filter(variable == "sptinc992j")
```

Merge the two Data Sets

```{r}
data50 <- DDCG %>%
  left_join(WID50, by = c("wbcode" = "iso3", "year" = "year")) %>%
  select(country_name, wbcode, year, dem, y, yeardem, top)
pdf50 <- pdata.frame(data50, index = c("wbcode", "year"))
```

Simple Model

```{r}
model50_1 <- plm(
  log(top) ~ dem,
  data = pdf50,
  model = "within",
  effect = "twoways"
  )
summary(model50_1)
```

= bad at everything